agentscope-ai/ReMe

Health v2 · maintenance only

Adoption not part of the score

3551 stars · 306 forks observed · 2026-10-06

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

ReMe is a Python memory management kit for AI agents that stores durable memory as ordinary Markdown files with frontmatter and wikilinks. It turns conversations and resources into a self-evolving, local-first personal knowledge base with BM25 and optional embedding-based retrieval.

Use cases

• give my ai agent long-term memory
• store agent memories as markdown files
• build a self-evolving knowledge base for agents
• retrieve relevant context from past conversations for an llm
• share a memory workspace across multiple agents
• manage agent memory locally instead of a vector database

When to choose

• you want agent memory stored as human-readable, editable Markdown files
• you prefer local-first storage with rebuildable indexes over opaque databases
• you need BM25 and wikilink-based retrieval without loading the whole knowledge base
• multiple agents should share and evolve one knowledge workspace

When to avoid

• you need a fully managed cloud memory service
• you require heavy vector-only semantic search at massive scale
• your stack is not Python 3.11+

Facets

library · maturity active

rag agent-framework search-engine llm-inference large-language-models developer-tools python cli self-hosted cross-platform agent-memory markdown-knowledge-base bm25 memory-management local-first wikilinks ai-agents retrieval-augmented-generation

Member repositories

For agents

markdown · JSON · MCP: product_card(name="agentscope-ai/ReMe")